Rare earth element abundances in hydrothermal fluids from the … · 2016-08-30 · Rare earth...

Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

Geochimica et Cosmochimica Acta 74 (2010) 5494–5513

Rare earth element abundances in hydrothermal fluids from theManus Basin, Papua New Guinea: Indicators of

sub-seafloor hydrothermal processes in back-arc basins

Paul R. Craddock a,*,1, Wolfgang Bach b, Jeffrey S. Seewald a, Olivier J. Rouxel a,c,Eoghan Reeves a, Margaret K. Tivey a

a Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, 360 Woods Hole Road,

Woods Hole, MA 02543, USAb University of Bremen, Geosciences Department, Klagenfurter Straße, 28359 Bremen, Germany

c Universite Europeene de Bretagne, Institut Universitaire Europeen de la Mer, 29280 Plouzane, France

Received 6 July 2009; accepted in revised form 14 June 2010; available online 14 July 2010

Abstract

Rare earth element (REE) concentrations are reported for a large suite of seafloor vent fluids from four hydrothermal sys-tems in the Manus back-arc basin (Vienna Woods, PACMANUS, DESMOS and SuSu Knolls vent areas). Sampled vent flu-ids show a wide range of absolute REE concentrations and chondrite-normalized (REEN) distribution patterns (LaN/SmN � 0.6–11; LaN/YbN � 0.6 – 71; EuN=Eu�N � 1–55). REEN distribution patterns in different vent fluids range fromlight-REE enriched, to mid- and heavy-REE enriched, to flat, and have a range of positive Eu-anomalies. This heterogeneitycontrasts markedly with relatively uniform REEN distribution patterns of mid-ocean ridge hydrothermal fluids. In ManusBasin fluids, aqueous REE compositions do not inherit directly or show a clear relationship with the REE compositionsof primary crustal rocks with which hydrothermal fluids interact. These results suggest that the REEs are less sensitive indi-cators of primary crustal rock composition despite crustal rocks being the dominant source of REEs in submarine hydrother-mal fluids. In contrast, differences in aqueous REE compositions are consistently correlated with differences in fluid pH andligand (chloride, fluoride and sulfate) concentrations. Our results suggest that the REEs can be used as an indicator of the typeof magmatic acid volatile (i.e., presence of HF, SO2) degassing in submarine hydrothermal systems. Additional fluid data sug-gest that near-seafloor mixing between high-temperature hydrothermal fluid and locally entrained seawater at many ventareas in the Manus Basin causes anhydrite precipitation. Anhydrite effectively incorporates REE and likely affects measuredfluid REE concentrations, but does not affect their relative distributions.� 2010 Elsevier Ltd. All rights reserved.

1. INTRODUCTION

Rare earth element (REE) distributions in submarinehydrothermal fluids have been used extensively as a tracerof sub-seafloor processes associated with hydrothermal

0016-7037/$ - see front matter � 2010 Elsevier Ltd. All rights reserved.

doi:10.1016/j.gca.2010.07.003

* Corresponding author. Tel.: +1 773 834 3997.E-mail address: [email protected] (P.R. Craddock).

1 Present address: Origins Laboratory, Department of Geophys-ical Sciences, University of Chicago, 5734 South Ellis Avenue,Chicago, IL 60637, USA.

activity including crustal (i.e., aluminosilicate) alterationby high-temperature hydrothermal fluids. Studies of REEsin seafloor vent fluids have focused mostly on basalt-hostedhydrothermal systems along mid-ocean ridge (MOR)spreading centers where fluids have remarkably uniformchondrite-normalized REE (REEN) distribution patternscharacterized by a light-REE enrichment and large, positiveEu anomaly (Michard et al., 1983; Michard and Albarede,1986; Michard, 1989; Klinkhammer et al., 1994; Mitraet al., 1994; Bau and Dulski, 1999; Douville et al., 1999).REE enrichments in seafloor hydrothermal fluids relative

http://dx.doi.org/10.1016/j.gca.2010.07.003

mailto:[email protected]

http://dx.doi.org/016/j.gca.2010.07.003

Rare earth elements in hydrothermal fluids from the Manus Basin 5495

to ambient seawater reflect removal from crustal rock duringfluid–rock reaction. It has been suggested that plagioclasedissolution controls the distribution of REEs in submarinehydrothermal fluids because the chondrite-normalizedREE (REEN) compositions of MOR hydrothermal fluidsand plagioclase are similar (Campbell et al., 1988; Klink-hammer et al., 1994). Laboratory studies, however, suggestthat REE compositions of seafloor vent fluids are unrelatedto primary rock composition because REEN distributionpatterns in experimental hydrothermal solutions are differ-ent from primary REEN compositions of the volcanic rockor individual minerals with which these fluids have reacted(Bach and Irber, 1998; Moller, 2002; Allen and Seyfried,2005). These studies suggest that vent fluid REE composi-tions reflect solubility control during fluid–rock interactioninfluenced by aqueous REE speciation (c.f., Bau, 1991),which in turn is strongly influenced by numerous aspects offluid chemistry such as pH, temperature and the availabilityof ligands. Collectively, these factors lead to fractionation ofREEs in sub-seafloor environments.

This study reports REE data for 36 seafloor vent fluidsfrom four active hydrothermal systems (Vienna Woods,PACMANUS, DESMOS and SuSu Knolls) in the Manusback-arc basin, Papua New Guinea. Prior to this study, lim-ited data existed for REE compositions of vent fluids anddeposits (e.g., anhydrite) recovered from the Manus Basinor other back-arc basins (e.g., Douville et al., 1999; Bachet al., 2003). Bach et al. (2003) proposed that the heteroge-neous REEN patterns recorded by anhydrite recovered fromPACMANUS reflected differences in aqueous REE concen-trations that are associated with changes in fluid composi-tion (Cl�, F� and SO4

2� ligand concentrations) owing tovarying inputs of magmatic volatiles (H2O–CO2–SO2–HCl–HF). The Manus Basin is an ideal area to investigatefactors that govern the aqueous abundance of REEs becausethe composition of underlying crustal rocks differs signifi-cantly among vent fields, ranging from basalt to rhyolite(Sinton et al., 2003; Bach et al., 2007; Sun et al., 2007). Inaddition, vent fluid composition (e.g., pH and ligand concen-trations) and temperature differ substantially between fieldsowing to varying degrees of magmatic volatile degassingeither on-going or in the recent past (Seewald et al., 2006;Reeves, 2010). Our data are used to examine the relationshipbetween measured aqueous REE concentrations and hostrock compositions to assess whether aqueous REE abundancesare controlled primarily by crustal rock composition or as-pects of fluid chemistry that affect REE solubility duringsub-seafloor fluid–rock interaction. Following on from thestudy of Bach et al. (2003), our data are also used to examinethe sensitivity of aqueous REE abundances to different fluidcompositions owing to varying styles of magmatic acid vol-atile input. The present study provides a better understand-ing of how measured REE abundances in seafloor ventfluids can be used as tracers of sub-seafloor processes affect-ing the formation of submarine hydrothermal fluids.

2. GEOLOGIC SETTING

The Manus Basin in the Bismarck Sea, Papua NewGuinea (Fig. 1) is a rapidly-opening (�100 mm/year)

back-arc basin associated with subduction of the SolomonMicroplate beneath the New Britain arc (Taylor, 1979;Davies et al., 1987; Martinez and Taylor, 1996). Crustal exten-sion and spreading are complex and variable and occuralong several distinct lineations. Toward the west is theManus Spreading Center (MSC) bounded between the Wil-laumez and Djaul transform faults (Martinez and Taylor,1996). Lavas erupted at the MSC are dominantly basalticin composition (Both et al., 1986; Sinton et al., 2003; Bachet al., 2007). Several areas of hydrothermal activity havebeen identified in the MSC; Vienna Woods is the largestand most active of the known fields (Tufar, 1990; Tiveyet al., 2007). It is located slightly south of the major spread-ing center within an axial rift valley at a water depth of�2500 m.

To the east, the Eastern Manus Basin (EMB) isbounded by the Djaul and Weitin transform faults whererapid spreading is accommodated primarily by rifting andextension of existing crust (Martinez and Taylor, 1996).Lavas are erupted as a series of discrete en echelon neovol-canic ridges and volcanic cones of felsic (andesite to rhyo-lite) composition (Sinton et al., 2003; Bach et al., 2007).The arc-affinity of volcanic lavas (Sinton et al., 2003) isconsistent with the proximal location (<200 km) of theEMB to the actively subducting margin. The EMB hostsseveral known active hydrothermal systems. The PapuaNew Guinea–Australia–Canada–Manus (PACMANUS)hydrothermal system is located on the crest of the 35 kmlong, 500 m high Pual Ridge, between water depths of1650 and 1740 m (Binns and Scott, 1993). The ridge is con-structed of several sub-horizontal lava flows with composi-tions between andesite and dacite (Binns and Scott, 1993;Sinton et al., 2003). There are several discrete vent fieldswithin the PACMANUS system (Fig. 1B) that exhibitvarying styles of hydrothermal activity ranging from high-temperature (>300 �C) black smoker fluid venting fromsulfide-rich chimneys to low-temperature diffuse flows throughsediment and cracks in lavas. Further to the east, the DES-MOS and SuSu Knolls hydrothermal systems are locatedon individual volcanic structures in environments that aremarkedly different from the ridge-hosted Vienna Woodsand PACMANUS hydrothermal fields. DESMOS (OnsenField, Sakai et al., 1991; Gamo et al., 1997) is located atthe crescent-shaped North wall of a caldera located in awater depth of �1900–2000 m. Basaltic to andesitic pillowflows and hyaloclastite deposits arranged across several ter-races form the slopes of the caldera. Sedimentation andalteration of primary lavas is common and includes Fe-oxide staining and acid-sulfate (advanced-argillic) alter-ation, locally abundant native sulfur flows and extensivepresumed microbial mats (Sakai et al., 1991; Gamo et al.,1997; Gena et al., 2001). Further east, SuSu Knolls consistsof three discrete volcanic cones (Suzette, North Su andSouth Su; Fig. 1C) at water depths between �1140 and1510 m (Binns et al., 1997; Tivey et al., 2007). The NorthSu and South Su volcanic structures are composed of abun-dant porphyritic dacite flows showing variable acid-sulfatetype alteration (i.e., quartz–pyrophyllite–illite–alunite andnative sulfur) and sedimentation by mixed volcaniclasticand hydrothermal material (Binns et al., 1997; Yeats

Fig. 1. (A) Regional tectonic setting of the Manus back-arc basin showing major plates and active plate motions (solid gray arrows). Areas ofknown hydrothermal activity in the Manus Spreading Center (Vienna Woods) and Eastern Manus Basin (PACMANUS, DESMOS and SuSuKnolls) are indicated by white stars. (B) Spatial distribution of known vent deposits at PACMANUS. (C) Spatial distribution of knownhydrothermal deposits at SuSu Knolls. Seafloor bathymetry based on EM300 SeaBeam sonar (modified from Tivey et al., 2007).

5496 P.R. Craddock et al. / Geochimica et Cosmochimica Acta 74 (2010) 5494–5513

et al., 2000; Hrischeva et al., 2007; Tivey et al., 2007). TheSuzette volcanic structure is extensively coated in metallif-erous sediment and relict sulfide talus that overlies primaryand/or secondary volcanic features (Hrischeva et al., 2007;Tivey et al., 2007).

2.1. Hydrothermal activity

2.1.1. Vienna woods

Current hydrothermal activity is manifest as both fo-cused and diffuse fluid venting within an area of �150 m


by 100 m (Tufar, 1990; Tivey et al., 2007). Inactive sulfidechimneys extend across a total area �300 m by 100 m.Mildly acidic (pH (25 �C) 4.2–4.7), clear and gray-smokerfluids with temperatures between 273 and 283 �C were sam-pled exiting the tops of large sulfide-rich chimneys up to7 m in height (Seewald et al., 2006).

2.1.2. PACMANUS

Current hydrothermal activity occurs at several discretevent fields (Roman Ruins, Roger’s Ruins, Satanic Mills,Snowcap, Tsukushi and Fenway) that are each between50 and 200 m in diameter (Binns et al., 2007; Tivey et al.,2007). Vent fluids with a range of temperature and compo-sition were sampled from these fields, including high-tem-perature fluids (�300–358 �C) with focused dischargefrom black smoker chimneys, lower temperature white/gray-smoker fluids (150–290 �C) discharging from “dif-fuser” chimneys that lack a central open conduit, andlow-temperature diffuse fluids (<100 �C) exiting throughcracks in the volcanic basement or metalliferous depositsand sediments. The measured pH (25 �C) of all high-tem-perature vent fluids (>240 �C) are relatively acidic between2.3 and 2.8 (Seewald et al., 2006).

2.1.3. DESMOS

Hydrothermal activity occurs at the Onsen field (Gamoet al., 1997), which is a small (�30 m diameter) area ofmoderate temperature fluid discharge located along thenorthern interior slope of the DESMOS caldera. Seafloorventing is markedly different from high-temperature blacksmoker fluids and is manifest as thick, milky-white fluidsdischarging directly through extensively altered volcanicbreccia and hydrothermal sediments composed of abundantnative sulfur and anhydrite (Gamo et al., 1997; Seewaldet al., 2006; Bach et al., 2007). Despite the low-temperaturesof sampled fluids (<120 �C), measured pH (25 �C) valuesare low (<1.5) and aqueous sulfate abundances exceed thatof seawater concentrations (Seewald et al., 2006). On thebasis of aqueous compositions, these fluids have beentermed ‘acid-sulfate’ fluids (Gamo et al., 1997).

2.1.4. SuSu Knolls

At SuSu Knolls, hydrothermal activity and vent fluidcompositions are remarkably diverse (Seewald et al.,2006). At North Su, the summit of the volcano is domi-nated by a large sulfide-rich black smoker complex withindividual chimneys up to 11 m in height. Vent fluids sam-pled from this complex are similar to high-temperatureblack smoker fluids sampled from PACMANUS, with tem-peratures between 300 and 325 �C and moderate-to-low pH(25 �C) between 2.8 and 3.2 (Seewald et al., 2006). In con-trast, the flanks of the volcano are dominated by extensivedischarge of acid-sulfate fluids similar to that sampled fromDESMOS. acid-sulfate fluids from North Su have a milky-white color, lower temperatures (48 to 241 �C), are veryacidic (pH (25 �C)<1.8) (Seewald et al., 2006) and exitthrough cracks in hyaloclastite flows, extensively alteredvolcanic breccia and hydrothermal sediments composedof native sulfur and anhydrite (Yeats et al., 2000; Bachet al., 2007; Tivey et al., 2007).

There is currently more limited hydrothermal activity atSouth Su. This vent field is characterized by outcrops ofboth fresh and variably altered volcanics overlain by mostlyinactive sulfide chimneys and scattered oxide-stainedhydrothermal sediments (Yeats et al., 2000; Hrischevaet al., 2007; Tivey et al., 2007). In an area of diffuse ventingtoward the NW, extensively altered and bleached volcanicrocks were observed (Tivey et al., 2007). High-temperaturefluid venting from scattered diffuser-type chimneys was ob-served and sampled in two areas toward the S and SE. Themaximum measured temperature for these fluids was290 �C. The pH (25 �C) of these fluids is low, ranging from2.6 to 2.7 (Seewald et al., 2006).

The smaller structure of Suzette, located NW of NorthSu and South Su, is extensively covered by volcanic breccia,hydrothermal and hemipelagic sediment, mass-wasted sul-fide talus, Fe-oxide crusts and limited exposures of possiblehydrothermal stockwork (Hrischeva et al., 2007; Tiveyet al., 2007). The summit is characterized by large expansesof both relict and scattered active sulfide chimneys that arecommonly buried within thick sediment. Hydrothermalactivity is intermittent over broad sections of the Suzettemound. Five high-temperature vents, with fluids emanatingfrom sulfide-rich chimney edifices, were sampled. Tempera-tures ranged from 226 to 303 �C and measured pH (25 �C)from 3.5 to 3.8 (Seewald et al., 2006). A sixth fluid was sam-pled from a cracked pavement structure; this fluid had atemperature of �249 �C and a considerably lower pH(25 �C) of �2.3 (Seewald et al., 2006).

3. METHODS

3.1. Sample collection and processing

A total of 101 hydrothermal fluid samples were collectedfrom 36 vents during R/V Melville cruise MGLN06MV(July–September, 2006) using discrete samplers deployedby ROV Jason II. Samples were primarily collected in160 ml isobaric gas-tight samplers (Seewald et al., 2002)and supplemented with fluids collected in 755 ml Ti-syringeor “major” samplers (Von Damm et al., 1985). In mostcases, fluids were sampled in triplicate. Temperatures weremeasured with a thermocouple mounted directly on thesnorkel of the isobaric samplers, and in some cases, usingthe ROV temperature probe. Estimated uncertainty in themeasured temperatures is ±2 �C.

Sample aliquots for chemical analysis were extractedimmediately after recovery of the samplers at the end ofeach dive operation. Fluid samples were drawn into acidcleaned, high-density polyethylene (HDPE) Nalgenee bot-tles. Aliquots of fluid drawn for REE and other trace ele-ment analysis were immediately acidified to pH < 2 byaddition of Fisher Optimae grade HCl to prevent precipi-tation of metal sulfides and sulfates from solution duringstorage prior to on-shore analysis. In nearly all majorand gas-tight samplers, a precipitate (“dregs” fraction)formed on the interior walls and bottom of the sampleras the hydrothermal fluid cooled. The dregs were collectedon a 0.22 lm pore-size, 45 mm diameter Nylon filter byrinsing with high-purity acetone and Milli-Q water. The


Nylon filters were dried and stored in glass vials for on-shore processing. In addition, minor precipitates formedwithin several acidified aliquots during storage (referredto here as “bottle-filter” fraction). These precipitates wereseparated from the fluid by filtering through a 0.22 lmNuclepore� filter as part of shore-based sample processing.The concentrations of REEs in sampled fluids were deter-mined by separate measurement of element concentrationsin all dissolved, dregs and bottle-filter fractions followed bymathematical reconstitution of the original fluid. A massbalance for REEs indicates that for fluids analyzed in thisstudy, approximately 90% of REEs were in the dissolvedphase.

Dissolved fractions were prepared for analysis by gravi-metric dilution of a �0.20 g split of each solution using 5%Fisher Optimae grade HNO3. Sample dilution factors wereadjusted as a function of the chlorinity of the fluid to obtaina uniform 6 mM Cl concentration in all samples. This equa-ted to between 80 and 110 times dilution of the originalsample. Dilution to uniform Cl minimized the effect of var-iable matrix (primarily from Na+, the major cation in solu-tion) on analyte behavior within the plasma during analysisby inductively coupled plasma–mass spectrometry (ICP–MS).Dregs and bottle–filter fractions were treated separately.Particles were removed from filters into 30 ml Savillexe

vials by rinsing with 5 ml of concentrated Fisher Optimae

HNO3. The vials were sealed and placed on a hot plateovernight (�70 �C) to digest particles. The resulting solu-tions were then evaporated to dryness. The acid diges-tion/evaporation step was repeated a further two times toachieve complete dissolution of the sulfide–sulfate mix.The digested particles were redissolved and quantitativelydiluted in 5% HNO3 acid for analysis.

3.2. Analytical methods

3.2.1. Determination of REE concentrations

Analyses of REE concentrations were performed on aThermoElectron Element2 ICP–MS at the Woods HoleOceanographic Institution. Solutions were introduced intothe plasma using a Cetac Aridus� desolvating nebulizerto reduce isobaric interferences (e.g., 135Ba16O+ on151Eu+). Barium- and REE-oxide formation was monitoredthroughout the analytical session by periodic aspiration ofBa and Ce spikes. Barium-oxide formation was signifi-cantly less than 1%. Significant interference of BaO+ onEu+ would bias (decrease) the natural 151Eu/153Eu ratio(�0.916). In almost all samples no bias was observed andno correction for BaO interference was required. REE-oxide formation was typically less than 1–4% of totalREE concentration. Samples were spiked with 1 ppb 45Scand 115In internal standards to correct for fluctuations ofthe plasma during the analytical session. Unknown sampleconcentrations were calibrated against matrix-matched,multi-REE standards prepared from Specpure plasma solu-tion standards. Background intensities were measured peri-odically by aspirating 5% HNO3 blanks. Externalprecision, determined by triplicate analysis of randomly se-lected samples across multiple analytical sessions, was�10% (1r).

3.2.2. Calculation of vent fluid and end-member fluid

compositions

Typically, seafloor vent fluid compositions are reportedas endmembers assuming that the hydrothermal fluid con-tains no Mg owing to quantitative Mg uptake duringhigh-temperature fluid-rock interaction (Von Damm,1983; Von Damm et al., 1985). Nearly all sampled fluidsin this study contained measurable Mg. The presence ofnon-zero Mg reflects either the true non-zero Mg composi-tion of the fluid sampled at the seafloor (e.g., owing to near-seafloor mixing between seawater and hydrothermal fluid),or artifacts during sampling owing to entrainment of sea-water at the sampler snorkel. The measured Mg concentra-tion of fluids sampled in duplicate or triplicate at eachdiscrete vent orifice was typically near identical, which isinterpreted as reflecting the true Mg concentration of thefluid exiting the seafloor. Fluids sampled in this studyshould not be interpreted as being compromised by seawa-ter entrainment during sampling because it is unlikely thatidentical amounts of seawater (as indicated by the same Mgconcentration) would be entrained in two or more indepen-dently sampled fluids. REE concentrations for all vent flu-ids sampled at the seafloor are therefore reported at thelowest measured Mg concentration (Table 1).

REE concentrations in smoker-type fluids are also reportedas endmembers (Table 2) by extrapolation of replicate ventfluid compositions to zero Mg using least-squares linearregression forced to pass through the composition of ambi-ent seawater (Von Damm, 1983). The concept of a zero Mgend-member hydrothermal fluid is useful for understandingthe processes occurring in the sub-seafloor (e.g., high-tem-perature fluid–rock interaction) prior to mixing betweenhydrothermal fluid and seawater at or close to the seafloor.Calculation of end-member concentrations also allows directcomparison of species compositions among vent fluidsexamined in this and other studies.

REE concentrations in acid-sulfate fluids sampled fromDESMOS and SuSu Knolls (North Su) were not extrapo-lated to zero Mg. Our indications are that the concept of azero Mg end-member acid-sulfate hydrothermal fluid doesnot necessarily apply. Acid-sulfate fluids sampled in tripli-cate contain similar and consistently high Mg concentrations(P24 mmol/kg), but have elevated temperatures and verylow pH. In contrast to typical high-temperature black- andgray-smoker fluids, acid-sulfate fluids appear to absent ofconvecting seawater-derived hydrothermal fluids that reactextensively with fresh crustal rocks and instead appear tohave formed primarily by sub-seafloor mixing of a magmaticvolatile phase and Mg-rich seawater (Seewald et al., 2006)followed by reaction with extensively altered crustal rocks.

4. RESULTS

4.1. Rare earth element concentrations in Manus Basin

hydrothermal fluids

4.1.1. Manus Spreading Center (Vienna Woods)

Total REE concentrations (RREE) at Vienna Woodsrange from �2.6 to 5.8 nmol/kg (Table 1). Chondrite-normalized REEN distribution patterns are uniformly

Table 1Rare earth element concentrations (pmol/kg) in sampled smoker- and acid-sulfate-type fluids from the Manus Basin. Data are reported at the lowest measured Mg.

Sample Vent Site Temp(�C)

pH(25 �C)

Mg(mM)

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb RF/RCl(�1000)

RSO4/RCl(�1000)

RCa/RCl

Smoker fluids

VW1 Vienna Woods 282 4.4 1.4 1700 1750 180 532 88 738 81 14 43 14 30 21 0.03 0.9 0.116VW2 Vienna Woods 273 4.2 1.0 1021 1244 157 467 81 629 61 10 33 14 25 19 0.03 0.6 0.117VW3 Vienna Woods 285 4.7 1.1 740 958 110 291 45 390 32 7 20 19 19 11 0.03 1.5 0.105

RMR1 Roman Ruins 314 2.3 7.3 8493 10,417 837 2225 520 6848 187 27 132 21 57 65 0.19 1.6 0.030RMR2 Roman Ruins 272 2.3 15.9 979 2210 283 1134 278 2865 188 27 207 22 59 59 0.17 5.5 0.019RMR3 Roman Ruins 278 2.5 6.4 9585 15,412 1935 7369 1568 16,075 1121 134 626 86 221 170 0.22 2.5 0.034RMR4 Roman Ruins 341 2.6 3.6 12,500 22,500 2830 10,213 2075 11,182 1324 154 685 97 228 191 0.19 1.7 0.034RGR1 Roger’s Ruins 320 2.7 4.2 6500 11,332 1416 5926 1389 7698 959 102 438 55 135 85 0.23 2.7 0.040RGR2 Roger’s Ruins 274 2.6 8.6 1165 2225 230 868 250 1712 232 28 135 18 42 45 0.21 3.0 0.035SM1 Satanic Mills 295 2.6 8.2 1361 3326 480 2179 685 3417 545 73 357 59 143 120 0.29 3.7 0.022SM2 Satanic Mills 241 2.4 16.9 330 800 125 694 389 677 1583 395 2777 479 1307 965 0.63 15.6 0.013SM3 Satanic Mills 288 2.5 9.7 1078 2481 410 2137 1050 2477 1705 258 1362 221 593 443 0.40 5.3 0.026SC1 Snowcap 152 4.6 30.8 221 411 46 176 83 105 58 16 112 26 83 70 0.26 22.8 0.013SC2 Snowcap 180 3.4 24.2 352 827 70 284 56 140 50 12 70 16 40 40 0.32 10.3 0.013TK1 Tsukushi 62 5.7 44.4 309 661 73 293 80 185 70 17 95 28 75 55 0.13 38.5 0.022F1 Fenway 329 2.5 5.8 859 1971 232 1252 1868 7362 2073 283 1338 203 530 401 0.61 4.3 0.031F2 Fenway 343 2.7 4.7 19,716 33,980 3982 14,465 2851 8514 1979 250 1210 165 405 314 0.25 2.5 0.038F3 Fenway 358 2.7 4.5 15,000 27,172 3400 14,000 3400 7750 2750 320 1500 205 470 290 0.29 3.9 0.039F4 Fenway 284 2.4 8.9 3000 9130 1200 5260 1610 5060 1970 430 3100 385 1150 900 0.30 5.9 0.034F5 Fenway 78 5.0 44.7 1960 4700 560 2070 520 1075 655 110 760 nd 210 100 0.16 52.6 0.026

SZ1 Suzette 303 3.8 4.4 3478 5736 726 2607 444 2741 291 32 163 25 65 61 0.10 2.3 0.051SZ2 Suzette 274 3.6 8.6 3095 5250 624 2450 546 2701 373 41 192 27 76 66 0.12 2.9 0.066SZ3 Suzette 290 3.5 5.5 3035 6237 950 4652 1300 4798 841 88 347 49 110 83 0.12 1.5 0.066SZ4 Suzette 229 3.6 8.3 868 1767 256 1150 279 1607 214 31 125 42 53 41 0.10 2.4 0.059SZ5 Suzette 249 2.3 6.4 2400 4094 668 3979 2561 1304 4005 739 4637 864 2486 2076 0.79 6.7 0.040SZ6 Suzette 226 3.7 8.0 2322 4120 594 2549 606 5024 459 45 192 26 60 54 0.12 1.9 0.066NS3 North Su 300 3.4 1.6 10,689 18,408 2075 6913 840 3089 424 41 179 29 66 60 0.21 1.1 0.046NS5 North Su 299 3.2 7.4 836 1700 330 2031 752 2341 265 20 90 16 37 51 0.31 4.2 0.034NS6 North Su 325 2.8 1.6 590 690 99 470 282 1502 399 59 318 54 142 147 0.48 1.4 0.052SS1 South Su 271 2.6 5.3 536 1250 126 452 201 181 212 57 521 91 235 164 0.56 3.3 0.045SS2 South Su 288 2.7 6.8 480 681 70 207 61 621 51 13 62 16 47 58 0.43 3.2 0.042

Acid-sulfate fluids

D1 DESMOS 117 1.0 46.0 17,862 54,166 7331 31,845 7830 2948 7283 1160 6980 1330 4044 3826 0.28 297.5 0.019D2 DESMOS 70 1.4 50.4 20,805 71,826 11,431 55,550 15,736 6526 14,321 2267 13,865 2595 7773 7336 0.01 123.7 0.024

NS1 North Su 48 1.8 49.0 10,923 35,507 4755 19,616 4699 1782 4452 718 4452 876 2793 2846 0.09 78.5 0.018NS2 North Su 215 0.9 38.8 25,597 98,058 16,244 83,537 25,511 8865 26,764 4323 26,644 5155 15,724 15,225 0.23 335.5 0.020NS4 North Su 241 1.5 23.5 12,657 47,812 6540 29,310 9834 3549 13,534 2355 15,301 3064 9287 8503 0.94 48.0 0.012

Rare

earthelem

ents

inh

ydro

therm

alfl

uid

sfro

mth

eM

anu

sB

asin5499

Table 2End-member (zero Mg) rare earth element concentrations (pmol/kg) in smoker fluids sampled in the Manus Basin.

Sample Vent Site Temp (�C) pH (25 �C) La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb

VW1 Vienna Woods 282 4.4 1700 1750 180 532 88 738 81 14 43 14 30 21VW2 Vienna Woods 273 4.2 1021 1244 157 467 81 629 61 10 33 14 25 19VW3 Vienna Woods 285 4.7 740 958 110 291 45 390 32 7 20 19 19 11

RMR1 Roman Ruins 314 2.3 9918 12,164 977 2599 607 7996 219 31 154 24 67 76RMR2 Roman Ruins 272 2.3 1392 3140 402 1612 395 4072 268 39 294 32 83 84RMR3 Roman Ruins 278 2.5 10,829 17,413 2186 8326 1772 18,162 1267 151 708 98 249 192RMR4 Roman Ruins 341 2.6 13,262 23,871 3002 10,835 2201 11,863 1405 164 727 103 242 202RGR1 Roger’s Ruins 320 2.7 7071 12,328 1540 6447 1511 8374 1044 111 476 60 147 92RGR2 Roger’s Ruins 274 2.6 1390 2653 274 1035 299 2042 277 34 161 21 51 53

SM1 Satanic Mills 295 2.6 1617 3950 570 2588 813 4058 647 87 424 69 170 143SM2 Satanic Mills 241 2.4 480 1163 182 1009 566 984 2303 575 4039 697 1901 1404SM3 Satanic Mills 288 2.5 1317 3031 501 2612 1283 3027 2084 315 1664 270 725 541SC1 Snowcap 152 4.6 499 928 103 396 186 237 130 35 251 58 188 158SC2 Snowcap 180 3.4 619 1454 123 500 99 246 88 21 123 28 70 70TK1 Tsukushi 62 5.7 1470 3148 349 1397 379 881 335 81 452 135 357 262F1 Fenway 329 2.5 959 2200 259 1397 2085 8217 2313 315 1493 226 592 448F2 Fenway 343 2.7 21,528 37,103 4348 15,794 3113 9296 2161 273 1321 180 442 343F3 Fenway 358 2.7 16,318 29,560 3699 15,230 3699 8431 2992 348 1632 223 511 315F4 Fenway 284 2.4 3609 10,984 1444 6328 1937 6087 2370 517 3729 463 1383 1083F5 Fenway 78 5.0 9925 23,799 2836 10,482 2633 5443 3317 557 3848 nd 1063 506

SZ1 Suzette 303 3.8 3777 6228 788 2831 482 2976 316 35 177 27 70 67SZ2 Suzette 274 3.6 3660 6209 738 2897 646 3194 441 48 227 32 90 78SZ3 Suzette 290 3.5 3368 6921 1054 5162 1442 5324 933 97 386 55 122 92SZ4 Suzette 229 3.6 1020 2077 301 1352 327 1888 251 36 147 50 63 48SZ5 Suzette 249 2.3 2711 4625 755 4495 2893 1473 4525 834 5239 976 2808 2345SZ6 Suzette 226 3.7 2711 4811 693 2976 708 5866 536 52 225 30 70 63NS3 North Su 300 3.4 10,689 18,408 2075 6913 840 3089 424 41 179 29 66 60NS5 North Su 299 3.2 964 1960 380 2342 867 2700 306 23 104 18 43 58NS6 North Su 325 2.8 590 690 99 470 282 1502 399 59 318 54 142 147SS1 South Su 271 2.6 592 1381 139 499 222 201 234 63 576 101 260 181SS2 South Su 288 2.7 547 775 80 236 69 708 58 15 71 18 53 66

5500P

.R.

Crad

do

cket

al./G

eoch

imica

etC

osm

och

imica

Acta

74(2010)

5494–5513


light-REE (La–Nd) enriched and show a large positiveEu-anomaly (Fig. 2). The data are consistent with REEcompositions previously reported for Vienna Woods hydro-thermal fluids (Douville et al., 1999). The range of REEconcentrations and relative distributions are similar tothose measured in high-temperature (P280 �C) vent fluidsfrom basalt-hosted, mid-ocean ridge (MOR) hydrothermalsites (Fig. 2).

Fig. 2. Chondrite-normalized REE distribution patterns of (A) mid-oceaand Atlantic Oceans. Data are from Michard et al. (1983), Michard (1989(1999); (B) black/gray-smoker fluids sampled from Vienna Woods, Man

A

C

Log

[REE

] Ch

on

dri

te-N

orm

aliz

edLo

g [R

EE] C

ho

nd

rite

-No

rmal

ized

Fig. 3. Chondrite-normalized REE patterns of black/gray-smoker fluidRuins-Roger’s Ruins; (B) Satanic Mills; (C) Fenway and (D) Snowcap a

4.1.2. Eastern Manus Basin (PACMANUS, DESMOS and

SuSu Knolls)

Most hydrothermal fluids sampled from the EasternManus Basin have high REE concentrations relative to flu-ids from Vienna Woods and overall exhibit a broad rangeof REEN distribution patterns. Total concentrations ofREEs at PACMANUS range from �1.5 to 88 nmol/kg(Table 1). The REEN distribution patterns of most sampled

n ridge hydrothermal fluids and modern seawater from the Pacific), Klinkhammer et al. (1994), Mitra et al. (1994) and Douville et al.us Spreading Center (this study).

B

D

s sampled from PACMANUS, Eastern Manus Basin. (A) Romannd Tsukushi.


fluids are variably light-REE enriched with a positive Euanomaly (Fig. 3). Two vent fluids, Satanic Mills (ventSM2) and Fenway (vent F1), have REEN distribution pat-terns that show heavy-REE (Dy–Yb) enrichments and po-sitive Eu anomalies (Fig. 3B and C).

Total REE concentrations in acid-sulfate fluids at DES-MOS vary from �147 to 236 nmol/kg (Table 1) and are sig-nificantly greater than those of sampled high-temperatureblack smoker fluids at Vienna Woods and PACMANUS(see also Douville et al., 1999), but are similar to those mea-sured in acid-sulfate fluids from continental geothermalenvironments (e.g., Valles Caldera, New Mexico (Michard,1989); Rotokawa and Waiotapu, New Zealand (Wood,2001)). Chondrite-normalized REEN distribution patternsare generally flat with a only slight convex-upward curva-ture and do not have a clear positive or negative Eu anom-aly (Fig. 4A). The flat REEN distributions patterns differmarkedly from those of black and gray-smoker fluids vent-ing from chimney edifices at other vent fields.

Total REE concentrations and REEN distribution pat-terns show a high degree of variability in vent fluids at SuSuKnolls (Fig. 4). Total REE concentrations in acid-sulfate flu-ids sampled from North Su range from �93 to 350 nmol/kg(Table 2) and exhibit REEN distribution patterns that are flatwith no clear Eu anomaly (Fig. 4A), similar to REEN distri-bution patterns in acid-sulfate fluids at DESMOS. Totalconcentrations of REEs in smoker fluids from North Su,South Su and Suzette range from �6.5 to 43 nmol/kg. TheREEN distribution patterns of most black/gray-smoker flu-

Fig. 4. Chondrite-normalized REE patterns of hydrothermal fluids samacid-sulfate fluids from DESMOS (open symbol) and North Su (filled symsmoker fluids from Suzette and (D) black/gray-smoker fluids from South

ids are light-REE enriched with pronounced positive Euanomalies (Fig. 4B–D). Some smoker fluids sampled fromSuzette (vent SZ5) and South Su (vent SS1) have REEN

distribution patterns that exhibit greater enrichment of hea-vy-REEs and smaller Eu anomalies relative to other smokerfluids sampled from the same vent field (Fig. 4B and D).

5. DISCUSSION

5.1. Controls on aqueous REE compositions of seafloor

hydrothermal fluids

Although there is consensus that the rare earth elements(REEs) in submarine hydrothermal fluids are derived prin-cipally from the oceanic crust during high-temperaturefluid–rock interaction, the primary processes that controlthe observed chondrite-normalized REE distribution pat-terns of seafloor vent fluids are poorly understood. It hasbeen hypothesized that REEN distribution patterns of ventfluids directly reflect the REE composition of crustal hostrocks or constituent minerals, such as plagioclase, subjectto alteration (Campbell et al., 1988; Klinkhammer et al.,1994). Alternatively, it has been proposed that REEN distri-bution patterns of vent fluids reflect equilibrium partition-ing between fluid and minerals, controlled primarily byfluid chemistry (e.g., pH, redox, availability of ligands),temperature, pressure and alteration mineralogy (Bau,1991; Bach and Irber, 1998; Bach et al., 2003; Allenand Seyfried, 2005). Previous studies reporting REE

pled from DESMOS and SuSu Knolls, Eastern Manus Basin. (A)bol); (B) black/gray-smoker fluids from North Su; (C) black/gray-Su.

Fig. 5. REE distribution patterns of: (A) bulk primary basalts(stippled field) and bulk primary dacites (gray field) sampledalong the Manus Spreading Center (MSC) and Eastern ManusBasin (EMB). Data are from Sinton et al. (2003). Also shown arecharacteristic REE patterns of constituent mineral phases inmafic (basalt) and felsic (dacite, rhyolite) igneous rocks (cpx isclinopyroxene, bt is biotite, plg is plagioclase and olv is olivine).Data are from Schnetzler and Philpotts (1970). Data arenormalized to average chondritic composition. (B) variablyaltered dacites sampled from the beneath PACMANUS, in theEMB. Solid and dashed lines indicate dacites showing incipient(chlorite-smectite) and extensive (“acid-sulfate-type”, illitepyro-phyllite-alunite) alteration, respectively. Data are from Beaudoinet al. (2007) and are normalized to average chondritic compo-sition. (C) bulk primary dacites (gray field) and highly-altereddacites (“acid-sulfate-type” alteration, shown in solid black lines)sampled at the DESMOS hydrothermal field, EMB. Data arenormalized to average chondritic composition. Also shown areREE pattern of the highly-altered dacites normalized to that ofthe primary dacites (dashed gray lines). REE data are from Genaet al. (2001). The REE patterns of primary and altered rocksrecord evidence for significant and variable REE mobilizationduring fluid-rock interaction, similar to that inferred from thefluid REE compositions reported in this study. During low pH,acid-sulfate-type alteration, the REEs appear to be mobilizedwithout significant fractionation among the lanthanides (C). Seetext for discussion.


compositions of submarine hydrothermal fluids have fo-cused largely on high-temperature, black smoker fluidssampled at mid-ocean ridge hydrothermal systems, wherethe range of rock-types encountered by circulating fluidsis limited (i.e., all are basalt-dominated) and the composi-tions of sampled fluids (e.g., pH, Mg, F, SO4

2� concentra-tions, chondrite-normalized REE patterns) are generallysimilar. These similar characteristics make it difficult toasses how REE solubility in submarine hydrothermal fluidsis affected by changes in fluid composition and whetheraqueous REE abundances reflect that of crustal rockREE compositions. In contrast, hydrothermal fluids sam-pled from the Manus back-arc basin have interacted withcrustal rocks of a wide range of composition (from basaltto dacite and rhyolite) and are remarkably different in termsof their pH, fluid composition and ligand and REE abun-dances. Accordingly, these vent fluids serve as an ideal casestudy by which to identify the key factors that affect REEsolubility at submarine hydrothermal fluids and how aque-ous REE abundances can be used as tracers of hydrother-mal processes.

5.1.1. Influence of host–rock REE composition

The range of REEN distribution patterns observed inManus Basin seafloor vent fluids do not reflect variationsin primary whole-rock REE abundances in basalts fromthe Manus Spreading Center (MSC) and dacites from theEastern Manus Basin (EMB; Fig. 5). For example, light-REE enriched patterns of Vienna Woods vent fluids arevery different to whole-rock REE distribution patterns ofprimary basalts erupted in the MSC (Fig. 5A). Similarly,the range of REEN distribution patterns measured in ventfluids from PACMANUS, DESMOS and SuSu Knolls isdifferent relative to the uniform whole-rock REE distribu-tion patterns of dacites erupted in the EMB (Fig. 5A).Moreover, the range of REEN distribution patterns mea-sured in Manus Basin vent fluids do not appear to recordthe REE composition of specific constituent minerals (seealso Schnetzler and Philpotts, 1970; Hanson, 1980). In par-ticular, the range of REE fluid compositions cannot be ex-plained as reflecting discriminate leaching of REEs fromplagioclase as previously suggested (Campbell et al., 1988;Klinkhammer et al., 1994; Douville et al., 1999). Additionalevidence suggesting that vent fluid REE abundances are notcontrolled by primary rock composition is provided byREE concentrations and REEN distribution patterns thatvary substantially among fluids samples within individualvent areas on spatial scales (<100–500 m) over which theprimary REE composition of volcanic rocks are uniform(Sinton et al., 2003; Miller et al., 2006). The fluid composi-tional data suggest that REEs are fractionated from pri-mary crustal rock REE compositions during fluid–rockinteraction owing to differences in aqueous REE solubility.This idea is supported by the mineral and chemical compo-sitions of altered crustal rocks in the Manus Basin thatvary significantly from chlorite–smectite (argillaceous) toillite–pyrophyllite–cristobalite–anhydrite (advanced-argillic)assemblages (Lackschewitz et al., 2004; Paulick et al.,2005; Paulick and Bach, 2006; Bach et al., 2007) and

Fig. 6. Total REE abundance versus pH for black smoker andacid-sulfate fluids sampled in the Manus Basin (this study). REEconcentrations in black smoker fluids from unsedimented, mid-ocean ridge hydrothermal systems (symbol +) and in acid-sulfatefluids from continental geothermal systems (symbol �) are shownfor comparison (data from Michard, 1989; Klinkhammer et al.,1994; Lewis et al., 1997; Douville et al., 1999). Smoker-type fluidswith lower temperatures (<280 �C) and elevated Mg concentrations(>10 mmol/kg) are not plotted as these fluids have experiencedlocal sub-seafloor mixing with seawater and the measured pH(25 �C) of these fluids does not reflect that of the high-temperaturehydrothermal fluid at depth.


indicate fluid–rock interaction involving a range of temper-atures (>220–360 �C), pH (<2 to >3.5), water/rock ratiosand fluid compositions. The REE compositions of variablyaltered rocks show a range of chondrite-normalized compo-sitions (Fig. 5B and C) that are consistent with variablemobilization and extensive fractionation of the REEs dur-ing crustal rock alteration (e.g., Gena et al., 2001; Beaudoinet al., 2007).

5.1.2. Influence of fluid composition – pH and ligand

concentration

High-temperature black smoker and acid-sulfate fluidsin the Manus Basin exhibit a wide range of pH and fluidcomposition (e.g., alkali metal, Mg, Cl, F and SO4 abun-dances) that reflect different styles of sub-seafloor fluid cir-culation, a range of conditions of fluid–rock interaction andvariable crustal rock compositions (Seewald et al., 2006;Craddock, 2009; Reeves, 2010). The compositions ofhigh-temperature black and gray-smoker fluids at ViennaWoods, PACMANUS and SuSu Knolls (e.g., Mg depletionand alkali metal enrichment relative to ambient seawater)reflect extensive reaction (buffering) between convectingseawater-derived hydrothermal fluids and fresh basalts inthe MSC and dacites in the EMB (Craddock, 2009). Thelow pH and elevated F concentrations of black smoker flu-ids in the EMB (PACMANUS and SuSu Knolls) relative tothose from the MSC (Vienna Woods) may reflect assimila-tion of magmatic acid volatiles (H2O–CO2–SO2–HCl–HF)with convecting hydrothermal fluids owing to the proximityof EMB to the active subducting arc. Acid-sulfate fluids atDESMOS and SuSu Knolls in the EMB are products offundamentally different styles of hydrothermal activityand are probable submarine equivalents of fumarolesassociated with subaerial volcanism (Giggenbach, 1992;Hedenquist et al., 1994). These fluids appear to have formedvia shallow mixing of magmatic acid volatiles with entrainedseawater in the absence of large-scale seawater convectivecirculation and reaction with fresh crustal rocks (Seewaldet al., 2006). The low pH (25 �C) < 1–1.5 and high sulfateconcentrations (28–149 mmol/kg) in DESMOS and SuSuKnolls acid-sulfate fluids relative to ambient seawater(Gamo et al., 1997; Seewald et al., 2006) likely reflect addi-tion of SO2-rich magmatic vapors to seawater and dispro-portionation of SO2 at decreasing temperatures to yieldsulfuric acid (e.g., Holland, 1965). The high concentrationsof Mg (P24 mmol/kg) and low concentrations of fluid-mo-bile elements (e.g., alkali metals) in acid-sulfate fluids are acritical difference relative to that of typical black smokerfluids (Craddock, 2009). These data suggest that acid-sul-fate fluid compositions are not governed primarily by reac-tion with fresh crustal rocks, but reflect interaction withpreviously and variably altered crustal rocks. This idea issupported by the occurrence of highly-altered rocks con-taining cristobalite, alunite, pyrophyllite and native sulfurin proximity to acid-sulfate fluids at the seafloor at DES-MOS and SuSu Knolls (Gamo et al., 1997; Bach et al.,2007) that are analogous to the low pH, advanced-argillicalteration observed in subaerial epithermal environments(Hemley and Jones, 1964; Hemley et al., 1969).

We propose that the range of REE abundances observedin Manus Basin hydrothermal fluids are controlled primar-ily by fluid composition (e.g., pH, availability of ligands)and alteration mineralogy, which regulate aqueous REEsolubility during sub-seafloor fluid–rock interaction. Over-all, REE concentrations in Manus Basin vent fluids increasewith acidity (Fig. 6). REE concentrations in acid-sulfatefluids from DESMOS and SuSu Knolls are high relativeto black and gray-smoker fluids sampled in the ManusBasin, despite the lower temperatures of acid-sulfate fluids.Similarly, REE concentrations are high in lower pH smokerfluids from PACMANUS and SuSu Knolls (pH (25 �C)2.3–3.5) relative to higher pH smoker fluids sampled fromVienna Woods (pH (25 �C) P 4.2). Similar correlations inhydrothermal fluids sampled from a range of geologic envi-ronments, including basalt-dominated mid-ocean ridge(MOR) and continental geothermal systems (Michard andAlbarede, 1986; Michard, 1989; Lewis et al., 1997; Douvilleet al., 1999; Wood, 2001) support this relationship. It is un-likely that REE abundances in seafloor hydrothermal fluidsare limited by their abundance in the host rock because theconcentrations of REEs in both fresh and altered oceaniccrustal rocks (e.g., Michard et al., 1983; Beaudoin et al.,2007) are several orders of magnitude higher than thosemeasured in submarine hydrothermal fluids (see Fig. 5).Indeed, acid-sulfate fluids have the highest REE concentra-tions, despite these fluids having bulk aqueous composi-tions consistent with reaction with altered crustal rocks,which typically contain REE abundances lower than thosein primary igneous lavas (e.g., Beaudoin et al., 2007).The results suggest total REE abundances in submarinehydrothermal fluids are solubility controlled during


high-temperature fluid–rock interaction, are strongly influ-enced by pH, and are largely independent of crustal–rockREE abundances.

Sampled vent fluids are also characterized by a range ofchondrite-normalized (REEN) patterns that vary systemat-ically with pH and the availability of aqueous ligands.These trends are readily apparent upon examination ofLaN/YbN and EuN=Eu�N ratios as a function of relative li-gand abundances (Fig. 7). The LaN/YbN ratio is a measureof the relative light-REE (La) versus heavy-REE (Yb)abundance and reflects chemical fractionation among thelight- and heavy-REEs. The EuN=Eu�N ratio (Eu anomaly)is a measure of the fractionation of Eu relative to neighbor-ing REEs (Sm and Gd), which can be related to the varyingoxidation state exhibited by Eu (II and III) compared to tri-valent REEs in high-temperature hydrothermal fluids(Sverjensky, 1984). High-temperature smoker fluids fromall Manus Basin vent areas exhibit a wide range of RF/RCl ratios and a limited range of low RSO4/RCl ratios,the latter being the result of sulfate removal by anhydriteprecipitation and reduction to H2S during fluid–rock inter-action. Manus Basin smoker fluids with low F and SO4

2�

Fig. 7. Correlations between aqueous [RF]/[RCl] ratios and chon-drite-normalized LaN/YbN (A) and EuN=Eu�N ratios (B) in blacksmoker and acid-sulfate fluids from the Manus Basin. Data for allvent areas are grouped by measured pH (25 �C). Smoker fluids withlower temperatures (<280 �C) and elevated Mg concentrations(>10 mmol/kg) are not plotted. Smoker fluids (pH > 2 to 4) showdecreasing LaN/YbN and EuN=Eu�N ratios (increasing heavy-REEabundances relative to light-REEs and Eu) with increasing [RF]/[RCl] ratios. Acid-sulfate fluids (pH < 2) show no clear differencesin LaN/YbN and EuN=Eu�N ratios for a wide range of [RF]/RCl]ratios. Differences in ligand abundances affect aqueous REEN

distributions likely owing to the formation of various REE-ligandcomplexes with different stabilities. See text for discussion.

concentrations, and low RF/RCl (<0.03 lM/mM) andRSO4/RCl ratios (�1 lM/mM) have a characteristic light-REE enrichment and large positive Eu-anomaly (highestLaN/YbN and EuN=Eu�N ratios; Fig. 7). In contrast, smokerfluids that have a range of high F concentrations and highRF/RCl ratios have only minor light-REE enrichments andlower LaN/YbN and EuN=Eu�N ratios. The most fluoride-rich sampled fluids with the highest RF/RCl ratiosapproaching �0.60–0.80 lM/mM (Satanic Mills SM2, Fen-way F1, Suzette SZ5; Table 1, Fig. 7) exhibit a distinctlight-REE depletion and heavy-REE enrichment (LaN/YbN� 1) and a small or absent Eu anomaly (EuN=Eu�N �1–2).

We interpret the range of light- versus heavy-REEenrichments observed in these fluids as reflecting differencesin relative solubilities of the light- versus heavy-REEs as adirect result of significant differences in pH, ligand concen-tration and, therefore, the relative abundances of REE-chloride and REE-fluoride complexes. The pronouncedlight-REE enrichment and positive Eu-anomaly in smokerfluids with chloride-rich, but fluoride- and sulfate-poorcompositions implies that REE-chloride complexation actsto enhance the solubility of the light-REEs and Eu relativeto that of the mid- and heavy-REEs during fluid–rock inter-action and secondary mineral formation. In contrast, themarked heavy-REE enrichment in fluoride-rich smokerfluids suggests that formation of REE-fluoride complexesenhances the solubility of the heavy-REEs relative tolight-REEs during fluid-rock interaction in the presenceof elevated concentrations of fluoride.

These interpretations are supported by the results of spe-cies distribution thermodynamic calculations that were car-ried out to assess the aqueous speciation/complexation ofREEs in representative sampled Manus hydrothermal fluidsat in situ temperature and pressure across a broad range offluid compositions (compositions of the seven fluids usedfor illustration in speciation calculations are given Table 3;results of calculations are presented in Fig. 8). Details of thethermodynamic calculations used in this study are docu-mented in full in the Supporting Online Material (SOM).Briefly, calculations were carried out using the SpecE8program, included as a module of the Geochemist’sWorkbench (v6.0) software package (Bethke, 1996). TheSpecE8 program computes species distributions, mineralsaturation states and gas fugacity in aqueous solutions atspecified temperature and pressure. The thermodynamicdatabase used in SpecE8 calculations was generated usingSUPCRT 92 (Johnson et al., 1992) for a pressure of50 MPa (500 bars) and temperatures between 0 and400 �C, using updates for aqueous formation constants aslisted in the SOM. The pH value of each fluid measuredat 25 �C was used together with aqueous compositionaldata to calculate pH at in situ temperature. The redox po-tential for each fluid was calculated based on measuredH2(aq) concentrations and the assumption of H2(aq)–H2Oequilibrium. Equilibrium between H2S and SO4

2� was sup-pressed because both species were measured in our hydro-thermal fluids and are in a state of non-equilibrium. REEspeciation in chloride-rich, but fluoride- (<20–170 lmol/kg) and sulfate-poor (<2 mmol/kg) smoker-type vent fluids

Tab

le3

Ch

emic

alco

mp

osi

tio

ns

of

rep

rese

nta

tive

hyd

roth

erm

alfl

uid

su

sed

inth

erm

od

ynam

icR

EE

spec

iati

on

calc

ula

tio

ns.

Sam

ple

Ven

tS

ite

Tem

pp

Hp

HM

gC

lN

aK

Ca

Ba

Al

Mn

Fe

SiO

2S

O4

FC

O2

H2S

H2

(�C

)(2

5�C

)(i

nsi

tu)

(mm

ol/

kg)

(mm

ol/

kg)

(mm

ol/

kg)

(mm

ol/

kg)

(mm

ol/

kg)

(lm

ol/

kg)

(lm

ol/

kg)

(lm

ol/

kg)

(lm

ol/

kg)

(mm

ol/

kg)

(mm

ol/

kg)

(lm

ol/

kg)

(mm

ol/

kg)

(mm

ol/

kg)

(lm

ol/

kg)

Sm

ok

erF

luid

s

VW

1V

ien

na

Wo

od

s28

24.

44.

81.

469

051

221

.280

.055

<10

349

159

15.3

0.8

214.

41.

442

RM

R4

Ro

man

Ru

ins

341

2.6

3.37

3.6

650

495

77.2

22.3

92<

1028

3064

6817

.80.

412

59.

56.

353

SM

1S

atan

icM

ills

295

2.6

3.25

8.2

519

407

60.5

12.5

20<

1021

4527

9012

.22.

116

718

18.

025

SM

2S

atan

icM

ills

241

2.4

3.2

16.9

455

374

38.6

5.9

16<

1018

5310

4512

.87.

128

711

24.

54.

3S

Z5

Su

zett

e24

92.

32.

76.

461

049

344

.324

.216

1824

339

6114

.13.

048

016

.14.

87.

4A

cid-S

ulf

ate

Flu

ids

D1

DE

SM

OS

117

1.0

1.2

46.0

492

391

8.3

9.4

<1

480

4012

400

8.1

147

137

23.1

0.0

4.0

NS

2N

ort

hS

u21

50.

91.

1838

.844

234

07.

88.

9n

d10

7581

3103

10.0

149

128

80.0

0.0

20


is governed primarily by REE-chloride and REE-oxyhy-droxide complexes of trivalent REEs and chloride com-plexes of divalent Eu (e.g., Vienna Woods VW1, RomanRuins RMR4, Satanic Mills SM1; Fig. 8). REE-fluoridecomplexes are present in lesser amounts and REE-sulfatecomplexes are present at relatively low abundance. The pre-dominance of divalent Eu versus trivalent Eu reflects thestability of Eu2+ under reducing conditions and elevatedtemperatures characteristic of sampled vent fluids (e.g.,Sverjensky, 1984). The differences in the relative abundanceof REE-chloride and REE-oxyhydroxide complexes resultprimarily from differences in fluid temperature and pH withREE-chloride complexes increasing in abundance at highertemperature and lower pH. The calculated REE speciationdoes not change significantly across different geologic set-tings and crustal compositions (e.g., basalt-dominatedVienna Woods versus dacite-dominated PACMANUS) be-cause the fluid compositions are similar.

REE species distribution in high-temperature vent fluidswith a range of elevated fluoride concentrations (up to�500 lmol/kg) is increasingly governed by the presence ofREE-fluoride complexes, which occur primarily at the ex-pense of REE-chloride complexes (e.g., Satanic MillsSM2, Suzette SZ5, Fig. 8). The light-REEs occur predomi-nantly as chloride complexes, but the heavy-REEs are pres-ent predominantly as fluoride complexes. The relativeabundance of REE-hydroxide and REE-sulfate complexesare low; similar to that predicted for other high-tempera-ture smoker fluids. Europium is again predicted to occurexclusively as a divalent chloride complex. Theoreticaland experimental estimates of the relative stabilities ofREE-chloride and REE-fluoride complexes across thelanthanide group (e.g., Wood, 1990; Haas et al., 1995;Gammons et al., 2002; Migdisov and Williams-Jones, 2007;Migdisov et al., 2009) are not without uncertainties and dis-crepancies (see documentation of thermodynamic data inSOM). However, these studies suggest that chloride com-plexes of the light-REEs and Eu2+ are more stable thanthose of the heavy-REEs at elevated temperature and pres-sure, whereas fluoride complexes possibly show the oppo-site behavior, consistent with our interpretation of naturalsamples. Fractionation of the REEs during fluid-rock inter-action and secondary mineral formation owing to differ-ences in complexation and solubility of the REEs is alsoconsistent with the results of experimental studies that havedemonstrated enhanced mobility of the light-REEs anddivalent Eu2+ during fluid–rock interaction with chloride-rich hydrothermal fluids at elevated temperature andpressure (e.g., Allen and Seyfried, 2005). No experimentalstudies have yet investigated the relative solubility andmobility of the REEs during fluid-rock interaction in thepresence of fluoride-rich hydrothermal fluids at elevatedtemperature and pressure.

Marked differences in fluid composition likely also gov-ern the REE distributions observed in acid-sulfate fluids,which are distinct to those of smoker fluids. All acid-sulfatefluids exhibit almost flat REEN distribution patterns withno Eu-anomalies (LaN/YbN � 1–2.5, EuN=Eu�N � 1), simi-lar to REEN patterns identified in other acid-sulfate fluidssampled in the Manus Basin (e.g., Douville et al., 1999).

Fig. 8. Predicted REE species distributions in high-temperature smoker fluids from the Manus Basin. (A) Fluid VW1 (Vienna Woods); (B)fluid RMR4 (Roman Ruins, PACMANUS); (C) fluid SM1 (Satanic Mills, PACMANUS); (D) fluid SM2 (Satanic Mills, PACMANUS); (E)fluid SZ5 (Suzette, SuSu Knolls); (F) fluid NS2 (North Su, SuSu Knolls) and (G) fluid D1 (DESMOS). REE-chloride, REE-fluoride and REE-sulfate complexes are shown by black squares, white triangles and gray diamonds, respectively. REE-oxyhydroxide complexes are shown bygray circles and free REE ions by black crosses. Fluid compositions used in calculations are reported in Tables 1 and 3.


We interpret the high REE concentrations and flat REEN

patterns of acid-sulfate fluids as reflecting primarily theinfluence of low pH, which enhances the aqueous solubilityof all REEs. The very low pH is likely the critical parametercontrolling the observed abundances of REEs in acid-sul-

fate fluids because observations of natural systems suggestthat the solubility of REEs during fluid-rock reaction in-creases significantly with acidity (Fig. 6) (e.g., Michard,1989; Lewis et al., 1998; Fulignati et al., 1999). In effect,acid-sulfate-type alteration occurring at very low pH, as is

Fig. 9. Fractional abundance of (A) fluoride present in solution asHF and Al-fluoride complexes and (B) REE present in solution asREE-fluoride complexes, versus measured fluid pH (25 �C) forhydrothermal fluids listed in Table 3. The dashed line in (A) is themodeled fractional abundance of HF relative to total fluoride in theabsence of Al-fluoride complexation (i.e., Al3 + removed fromfluid compositions for species distribution calculations). The valuesin parentheses in (B) are measured fluoride concentrations. At lowpH in acid-sulfate fluids, fluoride is effectively bound as HF and isunavailable to complex REEs. The presence or absence of Al-fluoride complexes does not affect predicted REE-fluoride complexabundances (dashed line follows modeled fluids). At higher pH insmoker fluids, proportionally more fluoride is dissociated andavailable to complex with REEs. The fraction of REE complexedwith fluoride is significant in fluoride-rich (P400 lM) hydrother-mal fluids at pH > 2.


observed at DESMOS and SuSu Knolls, effectively leachesand mobilizes REEs from host rocks without fractionationamong the lanthanides. Indeed, REEN distribution patternsof acid-sulfate fluids are similar to that of both primary andaltered felsic crustal rocks in the Eastern Manus Basin (e.g.,Gena et al., 2001) with which acid-sulfate fluids have likelyreacted (Fig. 5C).

Results of thermodynamic species distribution calcula-tions suggest that formation of aqueous REE complexesin acid-sulfate fluids appears to exert a relatively minor con-trol over the relative solubility of the REEs. REE speciesdistribution calculations suggest that at lower temperaturesrelevant to most acid-sulfate fluids (<120 �C) the REEs arepresent in solution predominantly as free trivalent ions(e.g., DESMOS D1, Fig. 8). Similar results were obtainedin REE species distribution calculations reported for acid-sulfate fluid by Douville et al. (1999) and Bach et al.(2003). Europium can occur as trivalent Eu reflecting thelower temperature and more oxidizing composition of thesefluids relative to typical high-temperature smoker fluids. Alow pH is critical for maintaining high concentrations ofREEs in solution as uncomplexed ions. The formation ofREE-sulfate complexes as predicted by species distributioncalculations (owing to high sulfate concentrations in acid-sulfate fluids up to 150 mmol/kg) may further enhanceREE solubility. At higher temperatures (>200 �C), thermo-dynamic calculations predict that REE-chloride complexesincreasingly contribute to REE speciation in acid-sulfatefluids (e.g., North Su NS2; Fig. 8). Differences in the rela-tive abundance of REE-sulfate versus REE-chloride com-plexes do not obviously affect overall REE abundances,suggesting that pH is the more important control. MeasuredREE abundances and REEN patterns are not obviouslyaffected by the wide range of fluoride concentrations inacid-sulfate fluids (Fig. 7), suggesting that REE-fluoridecomplexes do not contribute to aqueous REE speciationin these fluids. The lack of REE-fluoride complexation inacid-sulfate fluids (Figs. 8 and 9) likely reflects quantitativesequestration of fluoride as HF� at very low pH owing tothe weak acid behavior of hydrofluoric acid at elevatedtemperature (pKa �3.1 at 25 �C and 4.6 at 200 �C). Ithas also been suggested that competitive formation ofAl-fluoride complexes may inhibit REE-fluoride complexformation in acid-sulfate fluids (Gimeno Serrano et al.,2000). Indeed, acid-sulfate fluids sampled in the ManusBasin have very high concentrations of dissolved Al (Table 3)and Al-fluoride complexes and HF� constitute the entiretyof aqueous fluoride species in acid-sulfate fluids (Fig. 9).The presence or absence of Al-fluoride complexes, however,does not affect the predicted abundance of REE-fluoridecomplexes in acid-sulfate fluids because formation ofAl-fluoride complexes in acid-sulfate fluids occurs predom-inantly at the expense of HF�, not REE-fluoride complexes(Fig. 9).

In summary, the data presented suggest that REE solu-bility during fluid–rock interaction at elevated temperatureand pressure is critically dependent upon fluid composition– in particular pH and the concentration of aqueous REEligands. Different REE abundances and REEN distributionpatterns among smoker- and acid-sulfate-type fluids in the

Manus Basin likely reflect differences in REE solubility ow-ing to variations in the relative abundance and stability ofREE-chloride, fluoride and sulfate complexes as a functionof different temperature, pH and ligand concentration (e.g.,RF/RCl and RSO4/RCl ratios). The observed correlationssuggest that light-REE and Eu enrichments (large positiveLaN/YbN and EuN=Eu�N ratios) are favored in hydrothermalfluids with low fluoride and sulfate concentrations (low RF/RCl and RSO4/RCl ratios) in which REE-chloride com-plexes are the likely dominant REE species. In contrast,heavy-REE (and to a lesser degree mid-REE) enrichmentsare favored in hydrothermal fluids with high fluoride con-centrations (high RF/RCl ratios) in which REE-fluoridecomplexes are likely important REE species. In acid-sulfatefluids, enrichment of all REEs appears to be criticallydependent upon the very low pH and possibly, to a lesserextent, the presence of REE-sulfate and/or chloride com-plexes. Importantly, the apparent sensitivity of aqueousREE distributions to pH and Cl, F and SO4

2� concentra-tions implies that the REEs can be used as tracers for input


of fluoride- and sulfur-rich magmatic acid volatiles (H2O–CO2–SO2–HCl–HF) in seafloor hydrothermal fluids. Fur-ther, because minerals in hydrothermal vent deposits (e.g.,anhydrite) often record REE distributions which preservethat of the original hydrothermal fluid, the REEs may alsoprovide critical information about hydrothermal processesassociated with the formation of seafloor vent deposits,such as fluid pH, composition and the extent and type ofmagmatic acid volatile input, all of which impact metalbehavior and metal-sulfide mineralization.

5.1.3. Consequences of seawater entrainment and fluid mixing

Studies of MOR hydrothermal systems have demon-strated that entrainment and sub-seafloor mixing of seawa-ter with rising high-temperature hydrothermal fluids canaffect seafloor vent fluid compositions owing to mineral pre-cipitation and/or dissolution (Edmond et al., 1995; Tiveyet al., 1995). Anhydrite (CaSO4) precipitation is a commonproduct of sub-seafloor mixing between high-temperaturefluids and locally entrained seawater (Tivey et al., 1995;

Fig. 10. Compositions of black/gray-smoker fluids from PACMANUS.chloride-normalized end-member Ca versus end-member sulfate; and (D)Symbols as follows: Roman Ruins, gray diamonds; Satanic Mills, blackcomposition of ambient seawater is shown by the star. Normalization of chydrothermal fluids from individual vent areas resulting from sub-seafloo(zero Mg, zero sulfate) hydrothermal fluid (�300 �C) and seawater are sh(<280 �C) vent fluids have measurable Mg and sulfate owing to local ehydrothermal fluid (A). Measured sulfate, Ca and REE concentrations inby conservative mixing owing to removal during anhydrite (CaSO4) preprecipitation is indicated by the magnitude of the negative sulfate end-m(fluid F5) show excess sulfate, Ca and REE concentrations relative to thospreviously deposited anhydrite in the Fenway mound.

Mills et al., 1998; Tivey et al., 1998; Mills and Tivey,1999; Humphris and Bach, 2005) and can incorporate sig-nificant amounts of REEs (Mills and Elderfield, 1995;Humphris, 1998). Substantial amounts of anhydrite havebeen recovered from beneath several PACMANUS ventdeposits (Binns et al., 2007; Tivey et al., 2007). Thedown-hole distribution of anhydrite veins in Ocean DrillingProgram Leg 193 drill core from PACMANUS demon-strates widespread precipitation of anhydrite extending todepths >300 mbsf (Bach et al., 2003; Roberts et al., 2003).Sulfate–sulfur isotopic compositions of anhydrite atPACMANUS cluster around that of seawater sulfateconsistent with deposition of anhydrite in response tosub-seafloor mixing between entrained seawater andhigh-temperature hydrothermal fluid (Roberts et al., 2003;Craddock and Bach, 2010).

The range of chemical compositions of seafloor vent flu-ids sampled at PACMANUS is consistent with on-goinganhydrite precipitation at some vent areas (Seewald et al.,2006; Reeves, 2010). The highest-temperature (P340 �C)

(A) sulfate versus Mg; (B) chloride-normalized Ca versus Mg; (C)chloride-normalized end-member REE versus end-member sulfate.triangles; Fenway, light-gray squares; Snowcap, open circles. Thealcium to chloride removes differences in Ca concentrations amongr phase separation. Conservative mixing lines between end-memberown by the black dashed lines in each plot. Many lower temperaturentrainment and sub-seafloor mixing of seawater with end-membermany lower temperature fluids are, however, less than that predictedcipitation (fluids SC1, SC2, RMR2, SM2). The extent of anhydriteember anomaly. In contrast, low-temperature fluids from Fenway

e predicted by conservative mixing, suggesting current dissolution of


fluids sampled at PACMANUS have near zero Mg and sul-fate concentrations, whereas lower temperature (<280 �C)smoker fluids exhibit a range of higher Mg and sulfate con-centrations consistent with mixing between end-memberhigh-temperature hydrothermal fluids and locally entrainedMg- and sulfate-rich seawater (Reeves, 2010)(Table 1;Fig. 10). Measured sulfate concentrations in most lowertemperature fluids are, however, significantly less than thatpredicted by conservative mixing between end-memberhydrothermal fluid (zero Mg and SO4) and ambient seawa-ter causing measured sulfate concentrations to extrapolateto negative values at zero Mg. Although the value has notrue physical meaning, the magnitude of the negative end-member concentration represents a proxy for the extentof non-conservative sulfate removal during fluid mixing.Lower temperature fluids also exhibit Ca concentrationsless than that predicted by conservative mixing betweenend-member hydrothermal fluid and ambient seawater(Fig. 10B). The extent of non-conservative Ca removal iscorrelated with that of sulfate (Fig. 10C) and is explainedby anhydrite precipitation during near-seafloor mixing ofhigh-temperature hydrothermal fluids and locally entrainedseawater. The REEs exhibit non-conservative behavior sim-ilar to Ca during fluid mixing suggesting that anhydrite pre-cipitation has caused measurable removal of aqueous REEs(Fig. 10D). On the basis of the observed correlations andthe magnitude of the negative sulfate end-member concen-trations, on-going deposition of anhydrite appears mostprevalent at the Snowcap and Roman Ruins (PACMANUS)vent areas.

Measured Mg concentrations in low-temperature(�78 �C) fluids exiting diffusely from the seafloor at Fen-way (fluid F5) are near seawater concentration and are con-sistent with large-scale entrainment of seawater byascending high-temperature hydrothermal fluids withinthe Fenway mound (Table 1). Sulfate and Ca concentra-tions in these diffuse fluids are higher than that predictedby conservative mixing between parent high-temperaturehydrothermal fluids and seawater (Fig 10A and B), suggest-ing that local dissolution of previously deposited anhydriteis on-going in some fluids at Fenway in response to lowfluid temperatures that enhance anhydrite solubility(Bischoff and Seyfried, 1978). Dissolution of anhydrite yieldsend-member sulfate and Ca concentrations in diffuse fluidsthat are higher than in the highest-temperature black smo-ker fluids at zero Mg (Fig. 10C). End-member REE concen-trations in diffuse fluids at Fenway are also higher than inparent high-temperature vent fluids (Fig. 10D) and areinterpreted as reflecting addition of REEs to solution as aresult of anhydrite dissolution. It should be recognized thathigh end-member concentrations of sulfate, Ca and REEsin diffuse fluids are an artifact of extrapolation to zeroMg and should not be interpreted as indicating the actualexistence of a high-temperature fluid with the specified zeroMg composition. Anhydrite dissolution represents a poten-tial significant source of REEs because total REE concen-trations in massive anhydrites recovered at the seafloor atFenway reach values as high as 50 ppm and are enrichedby several orders of magnitude relative to total REEconcentrations in high-temperature hydrothermal fluids

(Craddock and Bach, 2010). The REEN patterns of high-and low-temperature hydrothermal fluids and of massiveanhydrites at Fenway are similar (Craddock and Bach,2010) indicating that dissolution of anhydrite does not,however, affect relative REEN distributions among thesampled Fenway vent fluids.

6. SUMMARY AND CONCLUSIONS

Rare earth element data are reported for a wide rangeof seafloor vent fluids sampled from four hydrothermalsystems in the Manus back-arc basin (Vienna Woods,PACMANUS, DESMOS and SuSu Knolls). Chondrite-normalized REEN distribution patterns of Manus vent flu-ids show a wide range of compositions, which contrastwith the uniform REEN distribution patterns of mid-oceanridge (MOR) hydrothermal fluids. These differences arecorrelated with differences in fluid composition, in particu-lar pH and ligand (chloride, fluoride and sulfate) abun-dances. Vent fluid REEN distribution patterns are notobviously correlated to differences in crustal rock REEcompositions. The results imply that REE compositionsof hydrothermal fluids are controlled predominantly byconditions of fluid–rock interaction (i.e., pH, temperature,availability of complexing ligands), which affect the rela-tive solubility and mobilization of the REEs. Our dataare supported by experimental studies that have docu-mented fractionation of the REEs during high-temperaturefluid–rock reaction (e.g., Bach and Irber, 1998; Allen andSeyfried, 2005).

In the Manus Basin, the range of REEN patterns ob-served in seafloor vent fluids is interpreted as reflectingheterogeneous solubility/mobility of the REEs during fluid–rock interaction owing to the formation of REE-chloride,REE-fluoride and REE-sulfate complexes with a range ofaqueous stabilities. The range of fluoride and sulfate con-centrations measured in Manus hydrothermal fluids isattributed to differing amounts and styles of magmatic acidvolatile (i.e., H2O–CO2–HCl–HF–SO2) degassing (Gamoet al., 1997; Douville et al., 1999; Seewald et al., 2006;Reeves, 2010). A significant implication of our results are thatREE distributions in seafloor hydrothermal fluids can beused as indicators of styles of magmatic degassing at depth.It has been proposed that the uniform REEN pattern ofMOR hydrothermal fluids (i.e., light-REE enrichment andpositive Eu anomaly) primarily reflects exchange of REEduring plagioclase recrystallization (Campbell et al., 1988;Klinkhammer et al., 1994; Douville et al., 1999). The resultsof this study, however, suggest that a light-REE enrichmentand positive Eu anomaly reflect enhanced mobility of thelight-REEs and Eu owing to the predominance of REE-chloride complexes in chloride-rich and fluoride- and sul-fate-poor MOR hydrothermal fluids. Local processes,including sub-surface fluid mixing and mineral depositionand remobilization, may also affect REE concentrationsof seafloor hydrothermal fluids. However, REEN distribu-tions of seafloor vent fluids are primarily affected by key as-pects of fluid composition (e.g., pH, ligand concentration)that affect REE speciation and mobility during fluid–rockinteraction at depth.


This detailed study of REE compositions of seafloorvent fluids indicates that REEs can provide critical infor-mation about fundamental sub-seafloor geochemical pro-cesses associated with hydrothermal activity, includingconditions of fluid–rock interaction and the extent andstyles of magmatic acid volatile degassing. A better under-standing of aqueous REE behavior offers important con-straints for interpreting REE signatures preserved in themineral record. Future studies of REE-bearing mineraldeposits can be used to gain insight about geochemical pro-cesses pertaining to hydrothermal vent deposit formationand are essential for inactive and relict systems where accessto hydrothermal fluids is precluded.

ACKNOWLEDGMENTS

The authors acknowledge the Captain and crew of the R/V Mel-

ville and the ROV Jason II technical group for a very successfulcruise and sample collection. P. Saccocia and E. Walsh providedgreat help during shipboard fluid sample processing. D. Schneiderand S. Birdwhistell (WHOI) are acknowledged for assistance dur-ing rare earth element analysis by ICP–MS. Reviews by A.A.Migdisov and an anonymous reviewer provided critical, construc-tive and thought-provoking comments that greatly aided in the dis-cussion and presentation of our research. The authors thank theAssociate Editor, J.-I. Ishibashi, for his editorial assistance andsupportive and insightful comments. This study received financialsupport from the Ocean Drilling Program Schlanger Fellowship(to P.R. Craddock), the WHOI Deep Ocean Exploration InstituteGraduate Fellowship (to E. Reeves) and NSF Grant OCE-0327448.

APPENDIX A. SUPPLEMENTARY DATA

Supplementary data associated with this article can befound, in the online version, at doi:10.1016/j.gca.2010.07.003.

REFERENCES

Allen D. E. and Seyfried, Jr., W. E. (2005) REE controls inultramafic hosted MOR hydrothermal systems: an experimentalstudy at elevated temperature and pressure. Geochim. Cosmo-

chim. Acta 69(3), 675–683.

Bach W. and Irber W. (1998) Rare earth element mobility in theoceanic lower sheeted dyke complex: evidence from geochem-ical data and leaching experiments. Chem. Geol. 151(1), 309–

326.

Bach W., Roberts S., Vanko D. A., Binns R. A., Yeats C. J.,Craddock P. R. and Humphris S. E. (2003) Controls of fluidchemistry and complexation on rare-earth element contents ofanhydrite from the Pacmanus subseafloor hydrothermal sys-tem, Manus Basin, Papua New Guinea. Miner. Deposita 38(8),

916–935.

Bach W., Tivey M. A., Seewald J. S., Tivey M. K., Craddock P. R.,Niedermeier D. and Yoerger D. (2007) Variable basementcomposition and magma degassing affecting hydrothermalsystems in the Eastern Manus Basin. EOS Trans. AGU. Fall

Meet. Suppl. 88(52). Abstract #V21D-0749.

Bau M. (1991) Rare-earth element mobility during hydrothermaland metamorphic fluid–rock interaction and the significance ofthe oxidation state of europium. Chem. Geol. 93(3–4), 219–230.

Bau M. and Dulski P. (1999) Comparing yttrium and rare earths inhydrothermal fluids from the Mid-Atlantic Ridge: implicationsfor Y and REE behaviour during near-vent mixing and for theY/Ho ratio of Proterozoic seawater. Chem. Geol. 155(1), 77–90.

Beaudoin Y., Scott S. D., Gorton M. P., Zajacz Z. and Halter W.(2007) Effects of hydrothermal alteration on Pb in the activePACMANUS hydrothermal field, ODP Leg 193, Manus Basin,Papua New Guinea: a LA-ICP-MS study. Geochim. Cosmo-

chim. Acta 71(17), 4256–4278.

Bethke C. M. (1996) Geochemical Reaction Modeling. OxfordUniversity Press.

Binns R. A., Barriga F. J. A. S. and Miller, D. J. (2007) Leg 193synthesis: anatomy of an active felsic-hosted hydrothermalsystem, eastern Manus Basin, Papua New Guinea. In Proceed-

ings of the Ocean Drilling Program, Scientific Results (eds. F. J.A. S. Barriga, R. A. Binns, D. J. Miller and P. M. Herzig), vol.

193. pp. 1–71. doi:10.2973/odp.proc.sr.193.201.2007 (OceanDrilling Program).

Binns R. A. and Scott S. D. (1993) Actively forming polymetallicsulfide deposits associated with felsic volcanic rocks in theeastern Manus back-arc basin, Papua New Guinea. Econ. Geol.

88, 2226–2236.

Binns R. A., Scott S. D., Gemmell J. B., Crook K. A. W., andShipboard Scientific Party (1997) The SuSu Knolls hydrother-mal field, eastern Manus Basin, Papua New Guinea. EOS

Trans. AGU. Fall Meet. Suppl. 78(52) (abstr. #V22E-02).Bischoff J. L. and Seyfried, Jr., W. E. (1978) Hydrothermal

chemistry of seawater from 25 to 350 �C. Am. J. Sci. 278, 838–

860.

Both R., Crook K., Taylor B., Brogan S., Chappell B., Frankel E.,Liu L., Sinton J. and Tiffin D. (1986) Hydrothermal chimneysand associated fauna in the Manus back-arc basin, Papua NewGuinea. EOS Trans. Am. Geophys. Union 67, 489.

Campbell A. C., Palmer M. R., Klinkhammer G. P., Bowers T. S.,Edmond J. M., Lawrence J. R., Casey J. F., Thompson G.,Humphris S. E. and Rona P. (1988) Chemistry of hot springs onthe Mid-Atlantic Ridge. Nature 335(6190), 514–519.

Craddock, P. R. 2009. Geochemical Tracers of Processes Affecting

the Formation of Seafloor Hydrothermal Fluids and Deposits in

the Manus Back-arc Basin. Ph. D. thesis, MassachusettsInstitute of Technology-Woods Hole Oceanographic Institu-tion, Woods Hole, MA, 370pp.

Craddock P. R. and Bach W. (2010) Insights to magmatic-hydrothermal processes in the Manus back-arc basin asrecorded by anhydrite. Geochim. Cosmochim. Acta 74(19),

5514–5536.

Davies H. L., Honza E., Tiffin D. L., Lock J., Okuda Y., Keene J.B., Murakami F. and Kisimoto K. (1987) Regional setting andstructure of the western Solomon Sea. Geo-Marine Lett. 7, 153–

160.

Douville E., Bienvenu P., Charlou J. L., Donval J. P., Fouquet Y.,Appriou P. and Gamo T. (1999) Yttrium and rare earthelements in fluids from various deep-sea hydrothermal systems.Geochim. Cosmochim. Acta 63(5), 627–643.

Edmond J. M., Campbell A. C., Palmer M. R., German C. R.,Klinkhammer G. P., Edmonds H. N., Elderfield H., ThompsonG., and Rona P. (1995) Time series studies of vent fluids fromthe TAG and MARK sites (1986, 1990) Mid-Atlantic Ridgeand a mechanism for Cu/Zn zonation in massive sulphideorebodies. In Hydrothermal Vents and Processes (eds. L. M.Parson, C. L. Walker, and D. R. Dixon), vol. 87. GeologicalSociety Special Publication. pp. 77–86.

Fulignati P., Gioncada A. and Sbrana A. (1999) Rare-earthelement (REE) behaviour in the alteration facies of the activemagmatic-hydrothermal system of Vulcano (Aeolian Islands,Italy). J. Volcanol. Geotherm. Res. 88(4), 325–342.



http://dx.doi.org/10.2973/odp.proc.sr.193.201.2007


Gammons C. H., Wood S. A., and Youning L. (2002). Complex-ation of the rare earth elements with aqueous chloride at 200 �Cand 300 �C and saturated water vapor pressure. In Water–Rock

Interaction, Ore Deposits and Environmental Geochemistry: A

Tribute to David A. Crerar. (eds. R. Hellmann and S. A. Wood),vol. 7. The Geochemical Society Special Publication. pp. 191–207.

Gamo T., Okamura K., Charlou J. L., Urabe T., Auzende J. M.,Ishibashi J.-I., Shitashima K., Chiba H., Binns R. A. and GenaK. (1997) Acidic and sulfate-rich hydrothermal fluids from theManus back-arc basin, Papua New Guinea. Geology 25(2), 139–

142.

Gena K., Mizuta T., Ishiyama D. and Urabe T. (2001) Acid-sulphate type alteration and mineralization in the DesmosCaldera, Manus back-arc basin, Papua New Guinea. Resour.

Geol. 51(1), 31–44.

Giggenbach W. F. (1992) Isotopic shifts in waters from geothermaland volcanic systems along convergent plate boundaries andtheir origin. Earth Planet. Sci. Lett. 113, 495–510.

Gimeno Serrano M. J., Auque Sanz L. F. and Nordstrom D. K.(2000) REE speciation in low-temperature acidic waters and thecompetitive effects of aluminum. Chem. Geol. 165(3–4), 167–

180.

Haas J. R., Shock E. L. and Sassani D. C. (1995) Rare earthelements in hydrothermal systems: estimates of standard partialmolal thermodynamic properties of aqueous complexes of therare earth elements at high pressures and temperatures.Geochim. Cosmochim. Acta 59(21), 4329–4350.

Hanson G. N. (1980) Rare earth elements in petrogenetic studies ofigneous systems. Annu. Rev. Earth Planet. Sci. 8(1), 371–406.

Hedenquist J. W., Aoki M. and Shinohara H. (1994) Flux ofvolatiles and ore-forming metals from the magmatic-hydro-thermal system of Satsuma Iwojima Volcano. Geology 22(7),

585–588.

Hemley J. J., Hostetler P. B., Gude A. J. and Mountjoy W. T.(1969) Some stability relations of alunite. Econ. Geol. 64, 599–

612.

Hemley J. J. and Jones W. R. (1964) Chemical aspects ofhydrothermal alteration with emphasis on hydrogen metaso-matism. Econ. Geol. 59, 538–569.

Holland H. D. (1965) Some applications of thermochemical data toproblems of ore deposits; [II] Mineral assemblages and thecomposition of ore forming fluids. Econ. Geol. 60(6), 1101–

1166.

Hrischeva E., Scott S. D. and Weston R. (2007) Metalliferoussediments associated with presently forming volcanogenicmassive sulfides: The SuSu Knolls hydrothermal field, EasternManus Basin, Papua New Guinea. Econ. Geol. 102, 55–74.

Humphris S. E. (1998) Rare earth element composition ofanhydrite: Implications for deposition and mobility within theTAG hydrothermal mound. In Proceedings of the Ocean

Drilling Program, Scientific Results (ed. P. M. Herzig, S. E.Humphris, D. J. Miller, and R. A. Zierenberg), vol. 158. OceanDrilling Program. pp. 143–159.

Humphris S. E. and Bach W. (2005) On the Sr isotope and REEcompositions of anhydrites from the TAG seafloor hydrother-mal system. Geochim. Cosmochim. Acta 69, 1511–1525.

Johnson J. W., Oelkers E. H. and Helgeson H. C. (1992) SUPCRT92: a software package for calculating the standard molalthermodynamic properties of minerals, gases, aqueous speciesand reactions from 1 to 5000 bar, and 0 to 1000 �C. Comput.

Geosci. 18(7), 899–947.

Klinkhammer G. P., Elderfield H., Edmond J. M. and Mitra A.(1994) Geochemical implications of rare earth element patternsin hydrothermal fluids from mid-ocean ridges. Geochim. Cos-

mochim. Acta 58(23), 5105–5113.

Lackschewitz K. S., Devey C. W., Stoffers P., Botz R., EisenhauerA., Kummetz M., Schmidt M. and Singer A. (2004) Mineral-ogical, geochemical and isotopic characteristics of hydrother-mal alteration processes in the active, submarine, felsic-hostedPACMANUS field, Manus Basin, Papua New Guinea. Geo-

chim. Cosmochim. Acta 68(21), 4405–4427.

Lewis A. J., Komninou A., Yardley B. W. D. and Palmer M. R.(1998) Rare earth element speciation in geothermal fluids fromYellowstone National Park, Wyoming, USA. Geochim. Cos-

mochim. Acta 62(4), 657–663.

Lewis A. J., Palmer M. R., Sturchio N. C. and Kemp A. J. (1997)The rare earth element geochemistry of acid-sulphate and acid-sulphate-chloride geothermal systems from YellowstoneNational Park, Wyoming, USA. Geochim. Cosmochim. Acta

61(4), 695–706.

Martinez F. and Taylor B. (1996) Backarc spreading, rifting, andmicroplate rotation, between transform faults in the ManusBasin. Marine Geophys. Res. 18, 203–224.

Michard A. (1989) Rare earth element systematics in hydrothermalfluids. Geochim. Cosmochim. Acta 53, 745–750.

Michard A. and Albarede F. (1986) The REE content of somehydrothermal fluids. Chem. Geol. 55(1–2), 51–60.

Michard G., Albarede F., Michard A., Minster J. F. and CharlouJ. L. (1983) Rare-earth elements and uranium in high-temper-ature solutions from East Pacific Rise hydrothermal vent field(13 �N). Nature 303, 795–797.

Migdisov A. A. and Williams-Jones A. E. (2007) An experimentalstudy of the solubility and speciation of NdF3 in F-bearingaqueous solutions. Geochim. Cosmochim. Acta 71, 3056–3069.

Migdisov A. A., Williams-Jones A. E. and Wagner T. (2009) Anexperimental study of the solubility and speciation of the rareearth elements (III) in fluoride- and chloride-bearing aqueoussolutions at temperatures up to 300 �C. Geochim. Cosmochim.

Acta 73, 7087–7109.

Miller D. J., Vanko D. A., and Paulick H. (2006). Data report:petrology and geochemistry of fresh, recent dacite lavas at PualRidge, Papua New Guinea, from an active, felsic-hosted seafloorhydrothermal system. In Proceedings of the Ocean Drilling

Program, Scientific Results (eds. F. J. A. S. Barriga, R. A. Binns,D. J. Miller, and P. M. Herzig), vol. 193. Ocean DrillingProgram. pp. 1–31. doi:10.2973/odp.proc.sr.193.208.2006.

Mills R. A. and Elderfield H. (1995) Rare earth element geochemistryof hydrothermal deposits from the active TAG Mound, 26 �NMid-Atlantic Ridge. Geochim. Cosmochim. Acta 59, 3511–3524.

Mills R. A., Teagle D. A. H. and Tivey, M. K. (1998) Fluid mixingand anhydrite precipitation within the TAG mound. InProceedings of the Ocean Drilling Program, Scientific Results

(eds. P. M. Herzig, S. E. Humphris, D. J. Miller, and R. A.Zierenberg), vol. 158. Ocean Drilling Program. pp. 119–127.

Mills R. A. and Tivey M. K. (1999) Sea water entrainment andfluid evolution within the TAG hydrothermal mound: Evidencefrom analyses of anhydrite. In Mid-Ocean Ridges: Dynamics of

Processes Associated with Creation of New Ocean Crust (eds. J.R. Cann, H. Elderfield and A. Laughton). Cambridge Univer-

sity Press, pp. 225–248.

Mitra A., Elderfield H. and Greaves M. J. (1994) Rare earthelements in submarine hydrothermal fluids and plumes from theMid-Atlantic Ridge. Marine Chem. 46(3), 217–235.

Moller P. (2002) The distribution of rare earth elements andyttrium in water–rock interactions: field observations andexperiments. In Water–Rock Interaction (eds. I. Stober and K.Bucher). Kluwer Academic Publishers, pp. 97–123.

Paulick H. and Bach W. (2006) Phyllosilicate alteration mineralassemblages in the Active Subsea-Floor Pacmanus Hydrother-mal System, Papua New Guinea, ODP Leg 193. Econ. Geol.

101, 633–650.



Paulick H., Herzig P. M., and Hoernes, S. (2005) Data report: acomprehensive geochemical, mineralogical and isotopic datasetof altered dacitic volcanic rocks from the subsurface of thePACMANUS hydrothemal field (ODP Leg 193). In Proceed-

ings of the Ocean Drilling Program, Scientific Results (eds. F. J.A. S. Barriga, R. A. Binns, D. J. Miller, and P. M. Herzig), vol.193. Ocean Drilling Program. pp. 1–18. doi:10.2973/odp.proc.sr.193.209.2005.

Reeves E. (2010) Laboratory and field-based investigations ofsubsurface geochemical processes in seafloor hydrothermalsystems. Ph. D. thesis, Massachusetts Institute of Technology-Woods Hole Oceanographic Institution, Woods Hole, MA.

Roberts S., Bach W., Binns R. A., Vanko D. A., Yeats C. J., TeagleD. A. H., Blacklock K., Blusztajn J. S., Boyce A. J. and CooperM. J. (2003) Contrasting evolution of hydrothermal fluids in thePACMANUS system, Manus Basin: The Sr and S isotopeevidence. Geology 31(9), 805–808.

Sakai H., Gamo T. and Scientific, Crew of Cruise KH-90-3 (1991).Hydrothermal activity in the eastern Manus Basin, BismarckSea: A brief report of the Hakuho-Maru Cruise KH-90-3. Ridge

Events 2, 39–51.Schnetzler C. C. and Philpotts J. A. (1970) Partition coefficients of

rare-earth elements between igneous matrix material and rock-forming mineral phenocrysts-II. Geochim. Cosmochim. Acta 34,

331–340.

Seewald J. S., Doherty K. W., Hammar T. R. and Liberatore S. P.(2002) A new gas-tight isobaric sampler for hydrothermalfluids. Deep Sea Res. Part I: Oceanograph. Res. Pap. 49(1), 189–

196.

Seewald J. S., Reeves E., Saccocia P., Rouxel O. J., Walsh E., PriceR. E., Tivey M., Bach W., and Tivey M. (2006). Water–rockreaction, substrate composition, magmatic degassing, andmixing as major factors controlling vent fluid compositions inManus Basin hydrothermal systems. EOS Trans. AGU. Fall

Meet. Suppl. 87(52) Abstract #B34A-02.Sinton J. M., Ford L. L., Chappell B. and McCulloch M. T. (2003)

Magma genesis and mantle heterogeneity in the Manus Back-Arc Basin, Papua New Guinea. J. Petrol. 44, 159–195.

Sun W. D., Binns R. A., Fan A. C., Kamenetsky V. S.,Wysoczanski R., Wei G. J., Hu Y. H. and Arculus R. J.(2007) Chlorine in submarine volcanic glasses from the easternManus basin. Geochim. Cosmochim. Acta 71, 1542–1552.

Sverjensky D. A. (1984) Europium redox equilibria in aqueoussolution. Earth Planet. Sci. Lett. 67, 70–78.

Taylor B. (1979) Bismarck Sea; evolution of a back-arc basin.Geology 7, 171–174.

Tivey M. A., Bach W., Seewald J. S., Tivey M. K., Vanko D. A.,and Shipboard Science and Technical Teams (2007) Cruisereport R/V Melville, MAGELLAN-06. In Hydrothermal Sys-

tems in the Eastern Manus Basin: Fluid Chemistry and Magnetic

Structures as Guides to Subseafloor Processes. Woods HoleOceanographic Institution. p. 67.

Tivey M. K., Humphris S. E., Thompson G., Hannington M. D.and Rona P. A. (1995) Deducing patterns of fluid flow andmixing within the TAG active hydrothermal mound usingmineralogical and geochemical data. J. Geophys. Res. 100(B7),

12527–12555.

Tivey M. K., Mills R. A., and Teagle D. A. H. (1998). Temperatureand salinity of fluid inclusions in anhydrite as indicators ofseawater entrainment and heating within the TAG activemound. In Proceedings of the Ocean Drilling Program, Scientific

Results (eds. P. M. Herzig, S. E. Humphris, D. J. Miller, and R.A. Zierenberg), vol. 158. Ocean Drilling Program. pp. 179–190.

Tufar W. (1990) Modern hydrothermal activity, formation ofcomplex massive sulfide deposits and associated vent commu-nities in the Manus back-arc basin (Bismarck Sea, Papua NewGuinea). Mitt. Osterr. Geol. Ges. 82, 183–210.

Von Damm K. L. (1983) Chemistry of submarine hydrothermalsolutions at 21� North, East Pacific Rise and Guaymas Basin,Gulf of California. Ph. D. thesis, Massachusetts Institute ofTechnology-Woods Hole Oceanographic Institution.

Von Damm K. L., Edmond J. M., Grant B., Measures C. I.,Walden B. and Weiss R. F. (1985) Chemistry of submarinehydrothermal solutions at 21 �N, East Pacific Rise. Geochim.

Cosmochim. Acta 49(11), 2197–2220.

Wood S. A. (1990) The aqueous geochemistry of the rare-earthelements and yttrium: 2. Theoretical predictions of speciation inhydrothermal solutions to 350 �C at saturation water vaporpressure. Chem. Geol. 88(1–2), 99–125.

Wood S. A. (2001) Behavior of rare earth elements in geothermalsystems: a new exploration/exploitation tool? US Departmentof Energy Report, University of Idaho, Moscow, ID, p. 35.

Yeats C. J., Binns R. A. and Parr J. M. (2000) Advanced argillicalteration associated with actively forming submarine polyme-tallic sulfide mineralization in the eastern Manus Basin, PapuaNew Guinea. Geol. Soc. Aust. Abstr. 59, 555.

Associate Editor: Jun-Ichiro Ishibashi



Rare earth element abundances in hydrothermal fluids from the … · 2016-08-30 · Rare earth...

Documents

Transcript of Rare earth element abundances in hydrothermal fluids from the … · 2016-08-30 · Rare earth...